期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2012
卷号：43
期号：1
页码：119-126
出版社：Journal of Theoretical and Applied
摘要：Due to highly complex of the grinding mechanism of the ball mill, it is a challenging problem to select the informative frequency spectral features of the response signals. High dimensionality and co-linearity of the frequency spectrum are unfavorable to build the effective mill load model in the wet ball mill. Interval Partial Least-Squares Regression (iPLS) is applied to select the feature frequency bands of the shell vibration signal and acoustical signal, which are closely relevant to the parameters of ball mill load. Redundant or irrelevant frequency spectral variables are removed to improve the complexity of ball mill load model and enhance the comprehension of the grinding process using the frequency spectrum features. The experimental results have demonstrated that the performance of the mill load models based on feature spectrum outperforms the full spectrum model for both the shell vibration signal and acoustical signal.
关键词：Ball Mill; Mill Load; Feature Selection; Interval Partial Least Square
